Emotion recognition in speech, driven by advances in neural network methodologies, has emerged as a pivotal domain in human–machine interaction. The deployment of sophisticated architectures such as ...

语音情感识别（Speech Emotion Recognition, SER）在过去基本遵循同一种范式：输入语音，输出情绪标签。这种设定在工程上有效，但在认知层面却过于简化。 在人类交流中，情绪判断从来不是一个 “标签选择” 的过程，而是一种基于证据整合的推理行为。我们会综合 ...